RNA Transfer from Poliovirus 135S Particles across Membranes Is Mediated by Long Umbilical Connectors Mike Strauss, Hazel C. Levy, Mihnea Bostina, David J. Filman and James M. Hogle J. Virol. 2013, 87(7):3903. DOI: 10.1128/JVI.03209-12. Published Ahead of Print 30 January 2013.
These include: REFERENCES
CONTENT ALERTS
This article cites 53 articles, 24 of which can be accessed free at: http://jvi.asm.org/content/87/7/3903#ref-list-1 Receive: RSS Feeds, eTOCs, free email alerts (when new articles cite this article), more»
Information about commercial reprint orders: http://journals.asm.org/site/misc/reprints.xhtml To subscribe to to another ASM Journal go to: http://journals.asm.org/site/subscriptions/
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
Updated information and services can be found at: http://jvi.asm.org/content/87/7/3903
RNA Transfer from Poliovirus 135S Particles across Membranes Is Mediated by Long Umbilical Connectors Mike Strauss,a Hazel C. Levy,a,b Mihnea Bostina,c David J. Filman,a James M. Hoglea Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts, USAa; Department of Biology, University of Florida, Gainesville, Florida, USAb; Facility for Electron Microscopy Research, McGill University, Montreal, QC, Canadac
T
he early steps leading to initiation of viral infection represent attractive targets for therapeutics and have been the subject of intensive interest for over 40 years. Studies of virus internalization mechanisms have played a major role in defining the diversity and complexity of the pathways used by the cell to internalize cargo (1, 2) but have often failed to produce information that would allow us to predict how any given virus is internalized. Thus, closely related viruses often use completely different endocytic pathways; the endocytic pathway used by a given virus is often different in different cell types, and some viruses are capable of using multiple pathways in a specific cell type (3, 4). In retrospect, this is not a surprise because it has become apparent that most of the machinery used for internalization is encoded by the cell and not by the virus, so it is arguably of some advantage for a virus to be flexible in the internalization pathways it uses. In contrast to flexibility in the mechanism that a virus uses to get into a cell, we postulate that virally encoded machinery plays a far greater role in the subsequent steps leading to the delivery of the viral genome or of a viral nucleoprotein complex to the correct compartment for replication. We believe that this machinery will be more highly conserved among closely related viruses, that aspects of the machinery will be conserved across more distantly related viruses, and that the features of this machinery will tell us about the mechanisms used by cells to translocate large cargoes across membranes. These hypotheses have already been substantiated for enveloped viruses, where fusion of the viral envelope with a cellular membrane provides a conceptually simple mechanism for delivery of the viral genome into the cytoplasm. Because nonenveloped viruses lack an external membrane, genome delivery requires that either the genome or a nucleoprotein complex containing the genome be translocated across a membrane to gain access to the inside of the cell (3–5). While the mechanism of genome translocation remains poorly understood for all nonenveloped viruses, poliovirus and its close relatives in the Enterovirus genus of the
April 2013 Volume 87 Number 7
Picornaviridae family have emerged as particularly attractive models for understanding one such process. The viruses are small (30 nm in diameter) and simple. Each virion is composed of 60 copies each of four capsid proteins (VP1, VP2, VP3, and VP4) that form a T⫽1 (pseudo-T⫽3) icosahedrally symmetric shell that encapsidates a positive-sense single-stranded RNA genome of 7,500 nucleotides (4). These viruses are highly amenable to biochemical, genetic, and structural characterization. Biochemistry of cell entry. Poliovirus infection is initiated when the virus attaches to a specific host cell surface receptor, called Pvr or CD155. This glycoprotein has an N-terminal ectodomain composed of three Ig-like domains, a transmembrane helix, and either of two splice-variant C-terminal intracellular domains (6). Only the first (most N-terminal) Ig-like domain contacts the virion (7–10). The ectodomain may be transferred to the anchor domain of other membrane proteins (11), or even to glycosylphosphatidylinositol (12), and remain a functional receptor. At physiological temperature, the receptor catalyzes (13) a conformational rearrangement of the poliovirus capsid, an expansion, producing the 135S or “A” particle (14, 15). This conformational change leads to the irreversible externalization of VP4 (which is myristoylated [16]) and of the N-terminal extension of VP1 (whose first 25 to 30 residues are predicted to be able to form an amphipathic helix in most enteroviruses [17]). Interestingly, these polypeptides have also been shown to be transiently and reversibly externalized when the virus is incubated at physiological temper-
Received 19 November 2012 Accepted 18 January 2013 Published ahead of print 30 January 2013 Address correspondence to James M. Hogle,
[email protected] Copyright © 2013, American Society for Microbiology. All Rights Reserved. doi:10.1128/JVI.03209-12
Journal of Virology
p. 3903–3914
jvi.asm.org
3903
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
During infection, the binding of poliovirus to its cell surface receptor at 37°C triggers an expansion of the virus in which internal polypeptides that bind to membranes are externalized. Subsequently, in a poorly understood process, the viral RNA genome is transferred directly across an endosomal membrane, and into the host cell cytoplasm, to initiate infection. Here, cryoelectron tomography demonstrates the results of 37°C warming of a poliovirus-receptor-liposome model complex that was produced using Ni-nitrilotriacetic acid lipids and His-tagged receptor ectodomains. In total, 651 subtomographic volumes were aligned, classified, and averaged to obtain detailed pictures, showing both the conversion of virus into its expanded form and the passage of RNA into intact liposomes. Unexpectedly, the virus and membrane surfaces were located ⬃50 Å apart, with the 5-fold axis tilted away from the perpendicular, and the solvent spaces between them were spanned by either one or two long “umbilical” density features that lie at an angle to the virus and membrane. The thinner connector, which sometimes appears alone, is 28 to 30 Å in diameter and has a footprint on the virus surface located close to either a 5-fold or a 3-fold axis. The broader connector has a footprint near the quasi-3-fold hole that opens upon virus expansion and is hypothesized to include RNA, shielded from enzymatic degradation by polypeptides that include the N-terminal extension of VP1 and capsid protein VP4. The implications of these observations for the mechanism of RNase-protected RNA transfer in picornaviruses are discussed.
Strauss et al.
generalized beta-barrel representation of VP1, VP2, and VP3 with the individual strands identified by letters. Loops are identified by the two beta strands that they connect (e.g., the BC loop of VP1 connects the B and C strands). (C) Internal network of intertwined VP4 and N-terminal polypeptides. (D) Channel at 5-fold axis. Numbers represent icosahedral symmetry axes.
ature, a process that has been dubbed “breathing” (18). The externalized polypeptides have been shown to associate with artificial membranes in vitro and with host cell membranes during the course of viral infection (17, 19, 20). Once the conversion to the 135S form has taken place, the particle is internalized by a noncanonical endocytic pathway (21). Shortly after internalization, while contained within a vesicle very near the cell surface, the virion releases its RNA directly into the cytoplasm, leaving behind an “empty” particle that sediments at 80S (21). The trigger for this release is unknown. Electrophysiology experiments have shown that 135S particles (or mature 160S virus particles maintained at 37°C to promote “breathing”) can induce the formation of channels, and ultimately pores, in planar membranes (12). Genetic studies have demonstrated that mutations in residue 28 of VP4 either abrogate the ability to form channels in vitro and the ability to release RNA during infection or alter the properties of the channels in vitro and delay the release of RNA during infection (19). This has led to the hypothesis that the insertion of VP4 and/or the N-terminal extension of VP1 into membranes facilitates the formation of pores in membranes that allow the translocation of the viral genome from the interior of the virus particle and into the cytoplasm of the cell. Virion structure. The structure of poliovirus has been determined at high resolution by X-ray crystallography (Fig. 1) (22– 24). The three major capsid proteins (VP2, VP3, and VP1) share a fold (a wedge-shaped -jellyroll) but have different sets of loops connecting the regular secondary structure elements of this core structure and have long N-terminal extensions that fold across the inner surface of the capsid. The outer surface of the virion is dominated by star-shaped mesas at the 5-fold axes and by three-bladed-propeller-shaped features surrounding the 3-fold axes. The star-shaped mesas are formed by loops that connect beta strands
3904
jvi.asm.org
at the narrow end of the VP1 beta barrel; the hubs of the propellers are formed by loops at the narrow ends of VP2 and VP3 as they alternate around the 3-fold axes. The blades of the propellers are contributed by the large EF loops of VP2, by the GH loop of VP1, and by the C terminus of VP2. (The loop naming convention is illustrated in Fig. 1B.) These prominent surface features are separated by canyon-like depressions surrounding the 5-fold mesas and by saddle-shaped depressions across each 2-fold axis. The canyon has been shown to be the receptor-binding site for polioviruses (7–10), rhinoviruses (25), and several other enteroviruses (4). The inner surface of the protein shell is decorated by an elaborate stabilizing network that is formed by the N-terminal extensions of VP1, VP2, and VP3 and by the small protein VP4 (Fig. 1C). At each 5-fold axis, there is a solvent-filled channel that is blocked by a plug that is formed on the inner surface of the capsid by intertwined N termini of VP3. The presence of this channel in the poliovirus and rhinovirus structures led to proposals that the channel (Fig. 1D) is the site of externalization of both viral peptides and genomic RNA during cell entry (26–28). Structures of cell entry intermediates. Cryo-electron microscopy (cryo-EM) structures of 135S particles (29, 30) and 80S particles (29, 31) showed that the particles had expanded by approximately 4% compared to 160S virus particles (29). Neither expanded structure was consistent with the proposals that RNA, VP4, and the N terminus of VP1 were released from a channel at the 5-fold axis. Indeed, this channel remains closed in both expanded virus structures. Instead, expansion causes both structures to have enlarged holes at their 2-fold axes and at the centers of their 5-3-3 icosahedral triangles (defined as having a 5-fold axis and two adjacent 3-fold axes at their corners). The center of the 5-3-3 triangle corresponds to the quasi-3-fold axis of T⫽3 viruses. Several factors, including mutational data and structural consid-
Journal of Virology
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
FIG 1 Virus structure. (A) Virion (left) and ribbon diagram of biological protomer (right; VP1, blue; VP2, yellow; VP3, red; VP4, green). (B) Schematic of the
Poliovirus RNA Translocation Membrane Complex
April 2013 Volume 87 Number 7
MATERIALS AND METHODS Virus purification. The Mahoney strain of type 1 poliovirus was grown in HeLa cell suspension culture. Cells were harvested by centrifugation, and virus was released from harvested cells by freeze-thaw lysis. Released virus was purified by differential centrifugation and CsCl density gradient as previously reported (37). Liposome preparation. Phosphatidylethanolamine, phosphatidylcholine, sphingomyelin, cholesterol, and phosphatidic acid in chloroform (Avanti Polar Lipids) were mixed in molar ratios of 1:1:1:1.5:0.3, respectively (17, 38), with nickel salt of lipids containing an NTA head group {1,2-dioleoyl-sn-glycero-3-[N-(5-amino-1-carboxypentyliminodiacetic acid)succinyl]} (Avanti Polar Lipids) added at a final concentration of 10% (wt/wt) (20, 35). The chloroform was evaporated under argon gas to produce a lipid film, which was dried under vacuum overnight. Dried lipid mix was solubilized in 50 mM HEPES (pH 7.3) and 50 mM NaCl to a final lipid concentration of 3 mg/ml. Liposomes were made by extruding rehydrated lipids through a membrane filter with 0.5-m-diameter pores (Avanti Polar Lipids). Virus-receptor-liposome complex formation. The extracellular domain of the soluble poliovirus receptor (sPVR) without the transmembrane and cytoplasmic domains but with a C-terminal six-histidine tag was obtained as a gift from V. R. Racaniello (Columbia University College of Physicians and Surgeons, New York, NY). Samples (0.1 g) of Histagged sPVR were added to the 50 l of liposomes and incubated at room temperature for 20 min. An 8-l volume of 1 mg/ml virus were added to receptor-decorated liposomes at room temperature before 8-l aliquots were each heated to 37°C for 4 min. After incubation, 4 l of sample was mixed with colloidal gold and placed on carbon-coated, Quantifoil holey carbon glow-discharged grids. To cryopreserve the samples, excess buffer was blotted from the grid and the grid was flash-plunged into liquid ethane with an FEI Vitrobot. Cryo-ET data. Virus-liposome complex tilt series were collected on a 300-kV Titan Krios (FEI Company, Hillsboro, OR) equipped with an UltraScan 4000 charge-coupled device camera (GATAN, Pleasanton, CA) over a tilt range of ⫾60° in 1.5° increments at a nominal magnification of ⫻35,000. The tilt series were processed and visualized using IMOD (39, 40), yielding tomograms with 3 Å or 4 Å per voxel. Two separate tomographic data sets were collected. The first included 394 particles obtained from 13 tomograms, and the second included 257 particles obtained from 15 tomograms (Fig. 2). Each virus particle was selected from a 4-fold binned tomogram and assigned two coordinates, one for the center of the virus and one for the closest contact point with the membrane, as estimated visually in IMOD (Fig. 2F). Virus particles were chosen only if they were located immediately adjacent to the membrane and well separated from other viruses. Icosahedral orientations. Using PEET (41), the capsid region of each chosen virus particle was initially centered and aligned to a low-passfiltered cryo-EM reconstruction of 135S particles (30). This determination of icosahedral orientations excluded the missing wedge of Fourier coefficients from calculations that correlated the Fourier transforms of observed and reference images (32, 36, 41). Radial masks were applied to both maps to increase the relative importance of the capsid region. At the current resolution, the general correctness of the icosahedral orientation parameters was verified by calculating icosahedral symmetrizations of the reoriented maps and looking for the characteristic appearance of peaks at the 5-fold and 3-fold axes and S-shaped depressions (having the correct handedness) across the 2-fold axes (see Fig. 1A). Symmetry breaking and calculation of reoriented subtomograms. All subsequent calculation steps were scripted by using SPARX (43) or BSOFT (44, 45). After establishing how the icosahedral symmetry elements were oriented in each subtomogram, a calculation was done for each chosen virus particle to determine which of the 60 icosahedrally equivalent rotations would cause the membrane contact vector (i.e., the vector between the two visually selected points) to lie within one specific 5-3-3 triangle. The 5-3-3 triangle (as diagrammed in Fig. 3B) covers 1/60
jvi.asm.org 3905
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
erations, have led to the suggestion that these holes represent the sites for release of viral polypeptides (30, 31), and a cryotomographic reconstruction of 80S particles caught in the act of the releasing RNA demonstrated that the viral RNA is released from a site somewhere between the 2-fold axis and the quasi-3-fold axis (32). Over the past several months, crystal structures have been reported for the 80S particle of HRV2 (at 3.0 Å resolution) (33) and for an 80S particle of enterovirus 71 (EV71; at 2.9 Å resolution) (34). Both structures show stunning similarities to our 80S cryo-EM reconstructions, including the observed holes at the 2-fold and quasi-3-fold axes, which reinforces the hypothesis that the nature of the structural transitions will be conserved among members of the Enterovirus genus. A simple membrane-containing model. Several years ago, we introduced a receptor-decorated liposome model (20, 35). In preparing the liposomes, a defined fraction of the lipids had Ni-nitrilotriacetic acid (NTA) head groups that later were used to capture ectodomains of the poliovirus receptor having C-terminal (membrane-proximal) His tags. We have previously demonstrated that these receptor-decorated liposomes capture virus, and we have solved the structure of the virus-receptor-liposome complex by cryo-EM (35) and cryo-electron tomography (cryo-ET) (36). Warming these complexes to physiological temperatures induces conformational changes in the virus that result in the insertion of both VP4 and the N-terminal extension of VP1 into the bilayer of the liposome (20). A specific proteolytic cleavage, at a site just proximal to an amphipathic helix at the VP1 N terminus, reverses the binding of virus to liposomes. This result demonstrates both that the amphipathic helix is essential for anchoring and that binding becomes receptor independent following virus expansion. We have recently shown that, upon heating to 37°C, this system is also capable of efficient translocation of the genome from the interior of the virus into the lumen of the liposomes, which remain intact (H. Levy et al. unpublished data). RNA translocation was resistant to large amounts of RNase added outside the liposomes, demonstrating that RNA is protected from RNase exposure during the entire translocation process in vitro. In vivo, parallel experiments have shown that the viral RNA is similarly protected from RNase (either added to the medium or tethered directly to virions) during infection. The failure of RNase to affect viral replication shows that the endosomal membrane remains intact, that genomic RNA is translocated across the intact membrane, and that it is protected from RNase at all stages of cell entry during natural infection. Here we report the use of this receptor-decorated liposome system to probe the structures of membrane complexes of altered particles that form when the virus-receptor-liposome complexes are incubated at 37°C. Using cryo-ET reconstructions of the complexes, we show that a substantial number of the virus particles appear to be captured in the act of RNA translocation. Surprisingly, all of the particles remain some distance from the membrane surface and most particles clearly show one or two stems of density (or umbilici) connecting the virus to the membrane. Asymmetric reconstructions obtained by averaging many individual subtomograms reveal that one of the umbilici intersects the virus surface near a quasi-3-fold axis, close to the site of externalization of viral polypeptides and of RNA in previous reconstructions of soluble forms of 80S particles.
Strauss et al.
dicular to the electron beam. Scale bar, 1,000 Å. (B) Gallery of individual subtomograms, oriented so that the membrane contact direction lies in the plane of the page. (C) Virion appearing to be partially engulfed by membrane. (D) Virions appearing to extrude RNA (arrow) through a membrane. (E) A single subtomogram with a visible connection point. (F) “Membrane contact vectors” were established visually, in three dimensions, by identifying the coordinates of virus centers (white circle) and nearby membrane contact points (black circle). Scale bar (B to F), 500 Å.
of the viral surface and has an icosahedral 5-fold axis and two neighboring 3-fold axes at its corners. Each virus (and its vicinity) was then reoriented by linear interpolation to occupy the center of a 256-by-256-by-256 map with a grid sampling of 3 Å per voxel, such that that the icosahedral symmetry elements of the capsids were placed into a standard orientation, and the visually identified membrane-contact vectors were oriented in as similar a direction as possible. A second set of these “reoriented subtomograms” was also produced in which the 5-2-3-2 quadrilateral (Fig. 3C) was used as the orientation target, rather than the 5-3-3 triangle, to allow artifact-free visualization of features close to the 5-3 edge. Averaging of oriented virus particles and visualization. To visualize a direct connection between virus and membrane, various subsets of the population of reoriented subtomograms were classified together on the basis of their physical attributes (see below), summed, low pass filtered to improve the signal-to-noise ratio, and visualized with Chimera (46), SPDBV (47), and COOT (48, 49). To aid in density interpretation, averaged or particle-specific maps were superimposed on pseudoatomic models that had previously been fitted to cryo-EM reconstructions of poliovirus 80S particles (3IYB.pdb, 3IYC.pdb [31]). Figures were prepared with Chimera (46).
RESULTS
Raw tomographic data. The raw tomographic data show the efficacy of the model system, where poliovirions are recruited to the
3906
jvi.asm.org
liposome surface by the ectodomain of the poliovirus receptor (Pvr), which in turn is attached to a lipid by a nickel-NTA tag. These virions remain attached to the liposomes after receptormediated conformational changes occur upon warming to 37°C. In many cases, several viruses are attached to a single liposome (Fig. 2). Some attachment events lead to membrane deformation, likely indicating that more than one connection exists between the liposome and the virus. In a number of the subtomograms, RNA can be seen in the process of exiting the virion and entering the liposome (Fig. 2D). Three-dimensional tomographic reconstruction. The raw tomographic reconstructions are noisy and suffer from the effects of the large “missing wedge,” a systematic absence of data in Fourier space. The signal-to-noise ratio can be improved, and distortions caused by the missing wedge can be minimized, by averaging over multiple subtomograms that have been oriented properly. A three-dimensional reconstruction obtained by averaging all 651 subtomograms clearly showed the virus particle and the nearby membrane. Surprisingly, the virus particle does not touch the membrane directly but is held approximately 50 Å away from the membrane surface. Density was clearly seen for two connec-
Journal of Virology
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
FIG 2 Tomographic reconstruction showing expanded poliovirions bound directly to liposomes. (A) Large field of view showing an 80-Å-thick slice perpen-
Poliovirus RNA Translocation Membrane Complex
FIG 3 Orientation of subtomograms for symmetric and subsequent asymmetric averaging. (A) An icosahedral average of 651 oriented subtomograms has large
tions (or umbilici), each about 55 to 60 Å long, that link the particle to the membrane (Fig. 4A). The locations of these connections and of nearby features on the viral surface can conveniently be described in the context of the triangular icosahedral unit (Fig. 3B), which has a 5-fold axis and two nearby 3-fold axes at its corners (the “5-3-3 triangle”). We will adopt the convention originally introduced by Rossmann (50) that describes the 5-fold axis as north and the 2-fold axis that lies midway between the two 3-fold axes as south. The larger of the two umbilical connections joins the virus particle in the vicinity of the quasi-3-fold axis, which lies at the center of the 5-3-3 triangle (Fig. 4A). This places the umbilicus above the canyon, nearly on a direct line between the 5-fold mesa and the tip of the nearest propeller blade (the EF loop of VP2). The smaller of the two umbilical connections joins the particle close to the 5-fold axis, to the north and west of the larger connection. Although the density level for both connections is significantly higher than the noise level, it is weaker than the density for the virus and the membrane. Additionally, the thickness of the reconstructed membrane is greater than expected, which suggests that there may be some variability in the membrane position among the particles used in the reconstruction. At higher contour levels, a distinct hole appears southeast of the larger (quasi-3-fold) connection. Neither the hole nor the connections are visible in any of the other 59 icosahedral triangles at a similarly high contour level. However, if the contour levels are dropped sufficiently, ghosts of the connections and of the hole appear in adjacent asymmetric units, which suggests that the data set includes a small percentage of misoriented particles. Preliminary classification. In order to explore possible sources of heterogeneity in the structures represented in the data, we evaluated many approaches to classifying the 651 properly oriented subtomograms prior to calculating various class averages. Meaningful differences between class averages were seen when the classification was based on “RNA content” or “emptiness,” as assessed by the ratio of solvent-subtracted density levels within the interior (RNA) to those in the capsid shell (protein). For this purpose, the densities in a radial shell from 90 Å to 170 Å were used to compute the average density for the shell (the radii were chosen on the basis of the structures of 80S and 135S particles), densities
April 2013 Volume 87 Number 7
from 0 Å to 90 Å were used to compute the average interior (RNA) density, and the voxels from 170 Å to 230 Å were used to compute the average solvent density. A histogram was plotted that showed the number of particles at each value of the “emptiness” ratio (Fig. 4F). That histogram showed an outlying group of 39 particles with anomalously low internal density levels. These may represent empty particles (which are present in any preparation of virus) that were never attached to a receptor. These empty particles were excluded from further analysis, and the remainder of the population was divided into tertiles by density ratio. Presumably, the particles in these three tertiles have externalized different amounts of their genomes. Meaningful differences, detailed below, were also seen after classification according to the distance between the membrane and the virus center, with the particles arbitrarily grouped into quartiles. In most classification attempts, the resulting class average reconstructions were remarkably similar to the average of all particles and included two umbilical connections that linked the membrane to the virus (Fig. 4C and D). In contrast, class averages for the lowest tertile of internal density and for the two quartiles with the largest particle-to-membrane distances lacked any visible connection when contoured at similar density levels (Fig. 4B and E). The lack of visible connections suggested that these particles were not linked to the membrane (either because they had never attached or because they had attached and subsequently detached), that the connection was too weak or flexible to be visible, or that these classes included a large percentage of misoriented particles. Paradoxically, the absence of either an umbilicus or a significant membrane distortion in these classes furnishes us with an important control. It confirms that the umbilicus is not merely the sum of missing-wedge artifacts (which cause individual boxed particle images to become elongated in the direction of the electron beam), and it confirms that the added density between virus and membrane is not merely an effect of low-pass filtering of a region that lies directly between two strong density features. “Best” averaged reconstruction. The three classes of particles that produced reconstructions that lacked visible connections were removed from the data set, and the oriented subtomograms from the remaining 206 particles were averaged to obtain an improved reconstruction (Fig. 5). As expected, this reconstruction is
jvi.asm.org 3907
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
outward projections at each 5-fold and 3-fold axis and S-shaped depressions or holes across each 2-fold axis. These are characteristic features of expanded virions and show that the subtomograms are well aligned. (B) The icosahedral 5-3-3 triangle and (C) the 5-2-3-2 quadrilateral are diagrammed in gray. These two areas, with symmetry axes at the corners, are two alternative choices for the icosahedrally unique 1/60 portion of the virus surface. Pentagons, triangles, and lens shapes indicate 5-fold, 3-fold, and 2-fold axes, respectively. (B) Purple and yellow patches, respectively, indicate the footprints of bound poliovirus receptor (7) and extruded RNA molecules, as previously seen in asymmetric reconstructions of expanded virions “caught in the-act” of extruding RNA (32). In the present experiment, particles were aligned with one another by choosing 1 of 60 icosahedrally equivalent orientations that would bring the membrane contact vector (as defined by the white and black circles in Fig. 2F) as close as possible to the center of the 5-3-3 triangle or the center of the 5-2-3-2 quadrilateral. Possible density artifacts near plane-figure boundaries were avoided by calculating every map in a way that features of interest were located near the center of the selected region.
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest FIG 4 Averages of subtomograms. Scale bar, 300 Å. For each class average, two orthogonal views are shown: In the face-on view, which is intended to show umbilicus locations on the capsid surface within the icosahedral triangle, the contour level was chosen with UCSF Chimera (46) so that the capsid exterior is shown as a mostly continuous solid surface and clipping planes section the umbilicus (hatched areas). The side view (rotated 90° about a vertical axis), with the membrane contact lying to the right of the virus, is clipped to make the umbilical connection between the virus and membrane as clear as possible, when it is present. To aid interpretability, two adjoining copies of the icosahedral 5-3-3 triangle are included. Prior to the calculation of each map, two possibly different icosahedral operators were applied to each subtomogram, one intended to rotate its membrane contact vector as close as possible to the center of the 5-3-3 triangle and the other intended to rotate its membrane contact vector as close as possible to the center of the 5-2-3-2 quadrilateral. To reduce calculation artifacts, we display the view that places the umbilical connection as far as possible from the edges of the plane figure. (A) The class average includes all 651 oriented subtomograms. Note that umbilicus locations (hatched areas) along the shoulder of the 5-fold axis and near the center of the 5-3-3 triangle are prevalent in the population. The third image from the left shows the unique hole in the capsid that appears at a higher contour level. The fourth image shows the locations of the umbilical features, relative to the position of the hole (red footprint), in the context of the icosahedral 5-3-3 triangle. (B, C) Density averages resulting from classification by particle “emptiness,” as measured by the ratio of solvent-corrected densities in the interior to those in the capsid. The least dense tertile of the population is averaged on the left (B), and the densest tertile is averaged on the right (C). Observe that visible umbilici are present in “fuller” particles but absent from “emptier” ones. (D, E) Density averages derived from classification by distance from membrane to virus center. The closest quartile (D) is averaged on the left, and the furthest quartile is averaged on the right (E). Note that no umbilicus is visible in the furthest class and no distortion of the membrane is apparent. We take this as an indication that some of the particles located near membranes are not actually attached and as evidence that the umbilicus is real (rather than an artifact of the calculation) when it is present. (F) Histograms show population distributions of virus-to-membrane distance, as measured from the virus center (left), and of density within the RNA region, relative to that in the capsid (right). Background shading indicates the portion of the population that was included in each of the class averages (B to E).
3908
jvi.asm.org
Journal of Virology
Poliovirus RNA Translocation Membrane Complex
highest RNA density and a proximate point of contact), the 5-fold and quasi-3-fold umbilici are both present at density levels similar to those of the capsid and membrane (middle). The membrane becomes visible at a lower contour level (right). As in the average of 651 images, the 5-fold axis of the virus is tilted, relative to the plane of the membrane. In contrast, the membrane is the expected width for a bilayer, which suggests a more conformationally homogeneous population. In panel A, crosshatching of the clipped umbilicus indicates density higher than that of the isocontour. Icosahedral symmetry elements are shown. Panels B and C are stereo views, without hatching, showing the 80S pseudoatomic model superimposed on the 206-particle map. In panel B, the atomic model helps to identify the positions of the umbilici on the virus surface. Note that the quasi-3-fold umbilicus is located above the “canyon,” midway between the 5-fold mesa and the propeller tip. (C) At a higher threshold, the unique hole in the capsid becomes visible, with Gln-68 of VP1 lying just below it (green ball).
generally similar in appearance to the all-particle reconstruction. It shows the virus to be linked to the membrane by two umbilical connectors, and it has a high-contour hole through the capsid, located near the quasi-3-fold axis, just southeast of the connector. These features are unique to the asymmetric unit that includes umbilical density, while the remainder of the virus particle appears to be approximately icosahedral. In contrast to the all-particle reconstruction, wherein the membrane appeared to be thicker than expected, and where the density for the connections was weak, in the 206-particle reconstruction, the thickness of the membrane is much closer to that expected for a bilayer. Furthermore, the densities for the two umbilical connections are comparable in strength to the density for the capsid and sometimes higher than the density level for the membrane (Fig. 5A).
April 2013 Volume 87 Number 7
The hole. Interestingly, the “hole” in the capsid (Fig. 4A) that appears at high contour lies immediately above Gln-68 in the Nterminal extension of VP1 from the adjacent protomer (Fig. 5C). This is the first residue of VP1 that is visible in the high-resolution structure of the 73S empty capsid assembly intermediate of poliovirus and also is the first well-ordered residue in the cryo-EM structures of the 80S particle (31) and the 135S particle (30). The residue lies at the “northernmost” end of the elongated hole spanning the 2-fold axes in 80S particles of poliovirus (31), rhinovirus 2 (33), and enterovirus 71 (34) and in the 135S particles of poliovirus (30) and coxsackievirus A16 (David Stuart, personal communication). In the high-resolution structure of the 135S particle of coxsackievirus A16, the N-terminal extension of VP1 makes a sharp turn near the corresponding residue and then heads radially
jvi.asm.org 3909
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
FIG 5 An improved average image results from exclusion of the emptiest and most distant particles. (A) In the average of the “best” 206 particles (having the
Strauss et al.
3910
jvi.asm.org
vidual subtomograms. In each case, the voxels to be included within the envelope were defined by the previous 206-particle unbiased average. Our initial objective was to find out whether individual virus particles could have both umbilici present simultaneously. This was to rule out the possibility that only one umbilical structure was present in each virus particle, with two possible locations appearing in the average of subtomograms. Our other objective was to find out whether any of the individual virus particles in the population had only one umbilicus or the other. This addressed the question of whether the umbilici could arise independently of one another, whether the formation of one specific umbilicus was dependent on the prior formation of the other, or whether the two umbilici were both components of a single structure and thus were never seen separately. To address all of these possibilities, we classified the individual virus particles into four groups, depending on whether the quasi3-fold umbilicus and the 5-fold umbilicus, respectively, had density levels greater (P ⫽ “plus”) or lesser (M ⫽ “minus”) than that of the capsid. Within the population of 206 “good” virus particles, the PP:MP:PM:MM proportions were 30%-20%-22%-27%. The uniformity of this distribution suggests that (except as described below) the two umbilici are indeed independent density features that can be present one at a time or both together. In contrast, among the 442 excluded particles, the proportions were 17%20%-20%-43%. Not unexpectedly, this confirms that excluding the emptiest and too-distant two-thirds of the population had the effect of eliminating a greater proportion of particles that had lower-level densities in the umbilical regions and of preferentially selecting for particles having two umbilici. In the hope of producing larger, more homogeneous populations and improved (albeit biased) reconstructions, we then restarted the classification process, eliminating only the 39 emptiest particles and the too-distant half, keeping the 309 particles represented by the major peak in the histogram of (RNA ⫺ Solvent)/ (Capsid ⫺ Solvent) (Fig. 4F). This population was then subdivided into four classes, depending on umbilical density levels relative to capsid, and subtomogram averages were calculated for all four of the classes (Fig. 6A to D). Not surprisingly, given the bias in the selection process, only the PP and PM class averages showed umbilical density features near the quasi-3-fold axis and only the PP and MP class averages had umbilici near the 5-fold axis (Fig. 6A and B). Nevertheless, four important results did come from the biased classification experiment. (i) In all three classes with umbilici, the distinctive hole in the capsid was present. (ii) In all three classes with umbilici, the distance and orientation of the virus, relative to the membrane, was the same, regardless of the number of visible umbilici present. (iii) In the MP class average, the 5-fold umbilicus was remarkably strong, sharp, and clear and roughly cylindrical, measuring 25 by 30 Å in cross-section and about 55 to 60 Å along its length. (iv) Most surprisingly, in the PM class average, the quasi-3-fold umbilicus was not alone, as selection against 5-fold umbilical density caused a previously unseen umbilicus to appear near the viral 3-fold axis (Fig. 6C). Presumably, this feature was not detected in earlier averages because it is present in less than 20% of the population. DISCUSSION
Previous models of RNA transfer. The original crystal structures of poliovirus and rhinovirus (22, 50) revealed that there are solvent-filled channels at the 5-fold axes that link the interior com-
Journal of Virology
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
to exit the particle at the site corresponding to the hole (David Stuart, personal communication). Within the 5-3-3 triangle, the unique quasi-3-fold hole is located just southeast of the footprint of the quasi-3-fold connector (Fig. 5B and C) and just west of the footprint of externalized RNA, as seen in the asymmetric reconstruction of 80S particles caught in the act of RNA release (Fig. 3B). The connections. The central connector lies approximately parallel to the quasi-3-fold axis of the virus particle. The apparent width of the connector at its narrowest point is ⬃25 to 30 Å. We estimate the distance between the virus surface and the membrane surface, measured along this connector, to be 35 to 60 Å, though the exact measurement clearly depends upon the choice of contour level, upon population homogeneity, and upon whether distance-based classification has been applied. Distances measured from membrane center to virus center (Fig. 4F) are somewhat less variable and average ⬃220 Å. The 5-fold connector intersects the virus particle just south of the 5-fold axis. At high contour (Fig. 5A, middle), it is apparent that the two connectors tilt away from one another as they leave the virus, resulting in a larger footprint on the membrane than on the capsid surface. Unlike the virus-receptor-liposome complex, where a 5-fold axis was oriented perpendicular to the membrane (35, 36), the membraneproximal 5-fold axis in the direct virus-membrane complex is tilted, bringing the unique quasi-3-fold axis closer to the membrane. Hypothetically, this rotation of the virus particle might be necessary in order to make both connections simultaneously, assuming that each has a specific length. When the “surface” representation of the reconstruction is viewed from above and the connections are cut by a forward clipping plane (Fig. 4A, C, and D and 5A and B), the footprints of the connectors appear as gaps in the isocontour surface and their sizes depend on the choice of contour level. Note, however, that the appearance of this gap is an artifact of the isocontour representation and that densities within the umbilicus are consistently higher than those outside. The footprints of the two umbilici are well separated from one another (Fig. 5B) and from the unique hole (Fig. 5C). (“Footprint” refers to the area on the virus surface that is covered by each of the umbilici and is approximated by the shape of the clipped density.) All three features are separated from the footprint of bound receptor (Fig. 3B), as seen in soluble virusreceptor complexes (7) and in the virus-receptor-liposome complex (35, 36). They are also distinctly different from the footprint of exiting RNA (Fig. 3B), as seen in asymmetric reconstructions of 80S particles caught in the act of releasing their RNA (32). The clipped isocontour surfaces are also useful for visualizing the locations of the footprints, relative to pseudoatomic models of the 80S structure (31). This approach (Fig. 5B and C) allows the positions of the connectors to be put in the context of known landmarks in the virus structure and, as discussed below, provides an indication of which polypeptide chains are likely to be involved in their formation. Biased classifications using density levels in the two umbilici. Initially, we calculated a number of unbiased averages that showed us two umbilical connections, one at the quasi-3-fold axis and the other near the 5-fold axis (Fig. 4 and 5). Once we had established that the umbilici were both real, we felt justified in performing a biased calculation that evaluates density levels (relative to the average capsid density) within a 5-fold-umbilical envelope and within a quasi-3-fold-umbilical envelope for each of the 651 indi-
Poliovirus RNA Translocation Membrane Complex
nondistant particles. Subtomograms were classified into four groups, based on whether their 5-fold and quasi-3-fold umbilici had density levels above or below the density of the capsid. (Left) Sections through the reconstructions (hatched areas) show the shape, size, and location of the umbilical features within the icosahedral triangle in face-on views. (Right) Corresponding side views. Panels: A, the PP class; B, the MP class; C, the PM class; D, the MM class. (A) The PP class average shows that particles with both 5-fold and quasi-3-fold umbilici must exist in the population. (B) The MP class average, with a strong 5-fold umbilicus but a weak quasi-3-fold umbilicus, shows strong, tubular density for the 5-fold umbilicus, 25 by 30 Å in cross-section and about 55 to 60 Å long. Note that the hole in the capsid near the unique quasi-3-fold axis is present at the same contour level as the umbilicus. This class demonstrates the possibility of the 5-fold umbilicus existing alone, in the absence of a quasi-3-fold umbilicus. (C) The PM class average, with a weak 5-fold umbilicus and a strong quasi-3fold umbilicus, unexpectedly, shows umbilical density near the icosahedral 3-fold axis. This class suggests the possibility that the quasi-3-fold umbilicus cannot exist unless a second umbilicus is also present. In the text, the hypothesis is discussed that the quasi-3-fold umbilicus includes viral RNA but the 5- and 3-fold umbilici do not, self-assembling from multiple copies of VP4 and N termini of VP1. (D) The MM class average, unsurprisingly, shows no evidence of a connection and may include particles that are near membrane but not attached.
April 2013 Volume 87 Number 7
jvi.asm.org 3911
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
FIG 6 A biased reclassification scheme was applied to the 309 nonempty,
partment of each virus with the exterior. The presence of these channels and their proximity to internal polypeptides that become externalized (Fig. 1C) led to a model of how the virus releases its RNA into the cell in the early stages of infection (42). This “5-foldeverything” model was both simple and attractive and was consistent with cryo-EM structures of the virus-receptor complex, both in soluble form (7) and in the virus-receptor-liposome complex (35, 36), that showed that receptor binding orients the virus such that one of its 5-fold axes points directly toward the membrane. However, more recent structural studies of poliovirus cell entry intermediates have shown that the N terminus of VP1 (and presumably VP4 as well) must be released through holes around the 2-fold and quasi-3-fold axes that become opened upon expansion of the virus in the 160S-to-135S transition. In particular, high-resolution structures of the expanded intermediates show that the VP3 plug remains in place and that the 5-fold channel remains too small to contain all of the externalized polypeptides (30, 31). Further evidence for the site of externalization comes from Fabs directed against the N terminus of VP1, which have been shown to localize to sites within the 2-fold depression in “breathing particles” (51), to bind at the tips of the EF loops of VP2 (“propeller tips”) in 135S particles (52), and to partition between the tips of the propeller blades and the 2-fold axes in 80S particles (52). Additionally, in high-resolution structures of expanded particles from various picornaviruses, the first ordered residue of VP1 (Gln-68) is consistently visible near the quasi-3fold hole (33, 34) (Fig. 5C). Finally, asymmetric reconstructions of 80S particles that were caught in the act of RNA release showed that RNA is released from the virus surface at a site close to the 2-fold and quasi-3-fold holes (32) (Fig. 3B). Although those structural results were clearly inconsistent with the original 5-foldeverything model of the release of polypeptides and RNA, the results did not immediately suggest an alternative model that was consistent with all of the data. What is the nature of the reconstructed particles? There are two possibilities for the nature of the membrane-bound particles that we see. If the particles represent either an early stage of 135S particle-membrane attachment or a very late stage, after RNA transfer has been completed, then the umbilical connections could be composed principally or exclusively of polypeptides. Alternatively, if RNA transfer has been initiated, but not completed, then an umbilicus could include single-stranded RNA in transit into the liposome, either alone or in complex with externalized peptides and/or lipids (see below). For a variety of reasons, we favor a model in which one of the connections (probably the connection near the quasi-3-fold axis) is composed of both RNA and peptides that protect the RNA from RNase digestion. (i) The level of density in the interior of the virus varies greatly (Fig. 4F), suggesting that many of the particles have externalized at least some of their viral RNA. A similarly variable level of internal density was observed in previously reported structures of the 80S “empty” particles of poliovirus and rhinoviruses (31, 53, 54) and was attributed to the repeated stalling of RNA release whenever stable secondary structures needed to be unwound (31, 32). In contrast, the RNA density within 135S particles is uniformly high (30). (ii) Fluorescence microscopy experiments, in which virus is bound to receptor-decorated liposomes that contain a dye that fluoresces when bound to RNA, showed that translocation of viral RNA into liposomes occurs efficiently under the conditions used to prepare the liposome complexes in the present structural studies (unpub-
Strauss et al.
3912
jvi.asm.org
ent study, the biased average reconstructions show the 5-fold umbilicus appearing alone but never show the quasi-3-fold umbilicus appearing alone. (v) Other than VP4 and the N-terminal extension of VP1, the polypeptides that become disordered in the 160Sto-135S transition (which therefore would be available for umbilicus assembly) are all much shorter and are located closer to the quasi-3-fold axis than to either the 5-fold or the 3-fold axis. The polypeptide segments that become disordered or rearranged in the 160S-to-135S transition and are located near the footprint of the quasi-3-fold umbilicus are shown in the superposition of the improved reconstruction and the pseudoatomic model (Fig. 5B). In particular, the GH loop of VP1 is seen to rearrange in the 160S-to-80S transitions (30, 31, 33, 34) and the C-terminal extensions of VP1 and VP2 clearly become disordered. Nearby, the GH loop of VP3 extends to form a two-stranded beta sheet (as seen in the 160S-to-80S transitions of EV71 [34] and poliovirus) and a third beta strand is observed to bind in the 135S structures of coxsackievirus A16 (David Stuart, personal communication) and poliovirus because of the anchoring of the externalized N-terminal extension of VP1. Any or all of these components could contribute to the formation of the quasi-3-fold umbilicus. Except for portions of the GH loop of VP1, none of these polypeptide segments is particularly hydrophobic or is predicted to associate with membranes. However, it is entirely possible that these components contribute to the structure of the quasi-3-fold connector and may help to provide protection for the RNA via protein-protein and protein-RNA interactions. Does the quasi-3-fold umbilicus function as a “helicase”? As discussed by Bostina et al. (32), both the slowness of RNA release and the variability of RNA density levels seen in individual RNAreleasing 80S particles are likely consequences of the need to unwind a series of RNA secondary structures (21). Hypothetically, membrane-induced formation of the quasi-3-fold umbilicus might well itself be the trigger for RNA unknotting and release. This is an attractive idea, as it would help to explain why the RNA inside 135S particles can remain folded stably at 37°C unless membranes are present, along with providing a ready explanation for how RNA consistently exits on the membrane-facing side of the virus. Symmetry breaking. Both the icosahedral symmetry of the virus particle and the 5-fold symmetry of the virus-membrane interaction are broken once one or more of the umbilici are formed and a pronounced hole forms in a unique icosahedral asymmetric unit. Despite this, the rest of the particle remains remarkably similar in structure to fully symmetric 135S and 80S particles. This suggests that changes within the unique 5-3-3 triangle might be slightly exaggerated versions of the symmetric changes that occur upon 135S and 80S particle formation in a variety of picornaviruses (30, 31, 33, 34). In each such expansion, a hinge-like motion flattens the curvature of the top surface of VP1, the intersubunit contacts within the biological protomer remain largely intact, and the interfaces between protomers are disrupted, causing holes to be opened at the 2-fold and quasi-3-fold axes. (The biological protomer is depicted in Fig. 1.) Perhaps an exaggerated hinge motion of VP1 and a concomitant expansion of the unique hole would create an opening of sufficient size both to allow passage of the RNA and to accommodate the other components of the connectors. How developing asymmetry in the complex may be coupled to the triggering of RNA transfer. Although we cannot rule out
Journal of Virology
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
lished results). (iii) A number of the raw tomograms show RNA in the process of being translocated across the liposome membrane. What components are available to form umbilici? In several of the class averages, there is a noteworthy consistency in the length and orientation of the umbilical density feature and in the virus position and orientation, as evidenced by the membrane density average having the expected width for a bilayer (Fig. 5A). If the umbilicus length or orientation were variable or if the umbilical structure were highly flexible, that would not be the case. This observation leads us to suspect that each umbilicus has a fixed composition and structure. There are several obvious candidates to consider for the components that might participate in umbilical structure formation. Receptors. For a variety of reasons, we feel that it is unlikely that the receptor is involved in forming the connectors. (i) The footprint of the receptor (Fig. 3B) is located some distance away from either connector. (ii) The affinity of an individual receptor for 135S and 80S particles is too low to measure (55). (iii) Even though the apparent affinity would be increased if multiple receptors were bound, the tilt of the particle with respect to the plane of the membrane (Fig. 4A) would make it nearly impossible for multiple receptors to remain bound. Membrane. Membrane involvement in the formation of umbilici is indicated by the failure of 135S particles to develop umbilici in the absence of membrane, by their formation on the membrane-facing side of the virus, and by their physical association with liposomes. N-terminal extension of VP1. Proteolytic data demonstrate that the first 30 amino acids at the N terminus of VP1 play a key role in membrane binding (probably mediated by an amphipathic helix formed by the first ⬃25 residues of VP1 that are conserved in all enteroviruses) (17, 20). Once this membrane association has occurred, each of the 70-residue-long N-terminal extensions would have an additional 30 to 40 residues available for the formation of umbilical structures. VP4. VP4 has been shown to interact with membranes in the liposome model (20) and during infection (19), and it has been shown to play a key role in genome translocation (19). In addition, a large number of copies of VP4 are released upon expansion that are free to diffuse both in solution and within the plane of the membrane. Together, these observations make it likely that multiple copies of VP4 are available to play a role in the formation of umbilical connections. Additional components of the quasi-3-fold connector. Five main factors suggest that the quasi-3-fold umbilicus, rather than the 5-fold and 3-fold umbilici, is directly involved in RNA transfer, providing the conduit for the passage of RNA from the virus to and across the liposome membrane. (i) The quasi-3-fold umbilicus is located closer to the footprint of externalized RNA (Fig. 3B) that was seen in asymmetric reconstructions of 80S particles that were caught in the act of RNA release (32). (ii) The quasi-3-fold umbilicus lies much closer to the unique hole seen at high contour levels in the present experiment in umbilicus-containing particles. (iii) The quasi-3-fold axis occurs in a deep depression on the virus surface, and the region includes interfaces between beta barrels that are known to separate during virus expansion, possibly providing a passage for the RNA. In contrast, the 5-fold (or 3-fold) umbilicus occurs high on the 5-fold mesa (or 3-fold propeller), where no major holes have previously been observed to open in either symmetrized or asymmetric maps (30–34). (iv) In the pres-
Poliovirus RNA Translocation Membrane Complex
4. 5. 6. 7.
8. 9.
10.
11. 12. 13. 14. 15. 16. 17. 18.
19. 20.
21. 22. 23.
ACKNOWLEDGMENTS This work was supported by grant NIH AI020566 (to J.M.H.). M.S. was supported in part by a fellowship from the Humboldt Foundation. We thank David Stuart for sharing data prior to publication.
24.
REFERENCES 1. Marsh M, Helenius A. 2006. Virus entry: open sesame. Cell 124:729 –740. 2. Mercer J, Helenius A. 2009. Virus entry by macropinocytosis. Nat. Cell Biol. 11:510 –520. 3. Levy H, Bostina M, Filman DJ, Hogle JM. 2010. Cell entry: a biochemical
April 2013 Volume 87 Number 7
25.
and structural perspective, p 87–107. In Ehrenfeld E, Domingo E, Roos R (ed), The picornaviruses. ASM Press, Washington, DC. Tuthill TJ, Groppelli E, Hogle JM, Rowlands DJ. 2010. Picornaviruses. Curr. Top. Microbiol. Immunol. 343:43– 89. Hogle JM. 2002. Poliovirus cell entry: common structural themes in viral cell entry pathways. Annu. Rev. Microbiol. 56:677–702. Mendelsohn CL, Wimmer E, Racaniello VR. 1989. Cellular receptor for poliovirus: molecular cloning, nucleotide sequence, and expression of a new member of the immunoglobulin superfamily. Cell 56:855– 865. Belnap DM, McDermott BM, Jr, Filman DJ, Cheng N, Trus BL, Zuccola HJ, Racaniello VR, Hogle JM, Steven AC. 2000. Three-dimensional structure of poliovirus receptor bound to poliovirus. Proc. Natl. Acad. Sci. U. S. A. 97:73–78. He Y, Bowman VD, Mueller S, Bator CM, Bella J, Peng X, Baker TS, Wimmer E, Kuhn RJ, Rossmann MG. 2000. Interaction of the poliovirus receptor with poliovirus. Proc. Natl. Acad. Sci. U. S. A. 97:79 – 84. He Y, Mueller S, Chipman PR, Bator CM, Peng X, Bowman VD, Mukhopadhyay S, Wimmer E, Kuhn RJ, Rossmann MG. 2003. Complexes of poliovirus serotypes with their common cellular receptor, CD155. J. Virol. 77:4827– 4835. Zhang P, Mueller S, Morais MC, Bator CM, Bowman VD, Hafenstein S, Wimmer E, Rossmann MG. 2008. Crystal structure of CD155 and electron microscopic studies of its complexes with polioviruses. Proc. Natl. Acad. Sci. U. S. A. 105:18284 –18289. Selinka H-C, Zibert A, Wimmer E. 1991. Poliovirus can enter and infect mammalian cells by way of an intercellular adhesion molecule 1 pathway. Proc. Natl. Acad. Sci. U. S. A. 88:3598 –3602. Tosteson MT, Wang H, Naumov A, Chow M. 2004. Poliovirus binding to its receptor in lipid bilayers results in particle-specific, temperaturesensitive channels. J. Gen. Virol. 85:1581–1589. Tsang SK, McDermott BM, Racaniello VR, Hogle JM. 2001. A kinetic analysis of the effect of poliovirus receptor on viral uncoating: the receptor as a catalyst. J. Virol. 75:4984 – 4989. Fenwick ML, Cooper PD. 1962. Early interactions between poliovirus and ERK cells: some observations on the nature and significance of the rejected particles. Virology 18:212–223. De Sena J, Mandel B. 1977. Studies on the in vitro uncoating of poliovirus. II. Characteristics of the membrane-modified particle. Virology 78: 554 –566. Chow M, Newman JFE, Filman D, Hogle JM, Rowlands DJ, Brown F. 1987. Myristylation of picornavirus capsid protein VP4 and its structural significance. Nature 327:482– 486. Fricks CE, Hogle JM. 1990. Cell-induced conformational change of poliovirus: externalization of the amino terminus of VP1 is responsible for liposome binding. J. Virol. 64:1934 –1945. Li Q, Yafal AG, Lee YM, Hogle J, Chow M. 1994. Poliovirus neutralization by antibodies to internal epitopes of VP4 and VP1 results from reversible exposure of these sequences at physiological temperature. J. Virol. 68:3965–3970. Danthi P, Tosteson M, Li QH, Chow M. 2003. Genome delivery and ion channel properties are altered in VP4 mutants of poliovirus. J. Virol. 77: 5266 –5274. Tuthill TJ, Bubeck D, Rowlands DJ, Hogle JM. 2006. Characterization of early steps in the poliovirus infection process: receptor-decorated liposomes induce conversion of the virus to membrane-anchored entryintermediate particles. J. Virol. 80:172–180. Brandenburg B, Lee LY, Lakadamyali M, Rust MJ, Zhuang X, Hogle JM. 2007. Imaging poliovirus entry in live cells. PLoS Biol. 5:e183. doi:10 .1371/journal.pbio.0050183. Hogle JM, Chow M, Filman DJ. 1985. Three-dimensional structure of poliovirus at 2.9 A resolution. Science 229:1358 –1365. Filman DJ, Syed R, Chow M, Macadam AJ, Minor PD, Hogle JM. 1989. Structural factors that control conformational transitions and serotype specificity in type 3 poliovirus. EMBO J. 8:1567–1579. Lentz KN, Smith AD, Geisler SC, Cox S, Buontempo P, Skelton A, DeMartino J, Rozhon E, Schwartz J, Girijavallabhan V, O’Connell J, Arnold E. 1997. Structure of poliovirus type 2 Lansing complexed with antiviral agent SCH48973: comparison of the structural and biological properties of three poliovirus serotypes. Structure 5:961–978. Olson NH, Kolatkar PR, Oliveira MA, Cheng RH, Greve JM, McClelland A, Baker TS, Rossmann MG. 1993. Structure of a human rhinovirus complexed with its receptor molecule. Proc. Natl. Acad. Sci. U. S. A. 90:507–511.
jvi.asm.org 3913
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
the possibility that RNA release occurs at a preexisting “weak spot” in the capsid, we are inclined to favor a stochastic model in which membrane proximity causes asymmetric structural changes in a previously symmetric complex and thereby creates the unique site for RNA translocation. In this model, several copies of the N terminus of VP1 could become externalized, perhaps while the virus is still attached to the receptor, by a process analogous to “breathing” of the virus at 37°C (18). As the amphipathic segments of some of these N-terminal extensions encounter membrane, they bind, thereby creating anchors that will persist as the virus dissociates from receptors. We envision the resulting tether as flexible and quite probably transient. At some point, irreversible conversion to a 135S-like intermediate should allow additional intermolecular contacts to form that reduce the flexibility of the tethers and rotate the virus so that a randomly chosen quasi-3-fold axis is brought into proximity to the membrane. This initiates the formation of the 5-fold (or 3-fold) umbilicus, composed of several membrane-anchored N-terminal extensions of VP1 and perhaps VP4. Once formed, this umbilicus holds the now-asymmetric complex rigidly at the observed distance and angle with respect to the membrane. Stresses associated with the formation of the 5-fold umbilicus, and with the concurrent tethering of the unique quasi-3-fold axis to the membrane, could then trigger further asymmetric changes in the capsid structure. These changes would result in the formation of the visible quasi-3-fold umbilicus, the opening of the unique hole, and the transport of RNA from the virus particle across the membrane via the quasi3-fold umbilicus. Such a sequence of events would be consistent with the variety of poliovirus-liposome class averages that we have seen. Note that the essential components of this model, including a long externalized N-terminal extension of VP1 with two conserved amphipathic regions, the externalization of the myristoylated protein VP4, and the nature of the conformational changes that the viruses undergo, are conserved among enteroviruses. Thus, the model is also likely to be applicable to other enteroviruses and perhaps more broadly in the Picornaviridae family. Conclusion. Many models have been proposed to explain how the RNA genome of a picornavirus is released from the virus and transferred to the cytoplasm to initiate infection (3, 5, 30, 42). However, during the past 10 years, a combination of structural and biochemical studies (including the present study) have contradicted all of the previous models. The receptor-decorated-liposome model system represents a very powerful tool, in that it allows virus structures to be determined in the context of membranes, and has begun to reveal the nature of the structural changes that are triggered by membrane attachment. This represents the first step in the generation of plausible models to replace the historical 5-fold-everything models. Like any good first step, it has generated many questions that should be addressable by future experiments.
Strauss et al.
3914
jvi.asm.org
39. Kremer JR, Mastronarde DN, McIntosh JR. 1996. Computer visualization of three-dimensional image data using IMOD. J. Struct. Biol. 116:71–76. 40. Mastronarde DN. 2008. Correction for non-perpendicularity of beam and tilt axis in tomographic reconstructions with the IMOD package. J. Microsc. 230:212–217. 41. Nicastro D, Schwartz C, Pierson J, Gaudette R, Porter ME, McIntosh JR. 2006. The molecular architecture of axonemes revealed by cryoelectron tomography. Science 313:944 –948. 42. Mosser AG, Sgro JY, Rueckert RR. 1994. Distribution of drug resistance mutations in type 3 poliovirus identifies three regions involved in uncoating functions. J. Virol. 68:8193– 8201. 43. Hohn M, Tang G, Goodyear G, Baldwin PR, Huang Z, Penczek PA, Yang C, Glaeser RM, Adams PD, Ludtke SJ. 2007. SPARX, a new environment for Cryo-EM image processing. J. Struct. Biol. 157:47–55. 44. Heymann JB. 2001. Bsoft: image and molecular processing in electron microscopy. J. Struct. Biol. 133:156 –169. 45. Heymann JB, Belnap DM. 2007. Bsoft: image processing and molecular modeling for electron microscopy. J. Struct. Biol. 157:3–18. 46. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. 2004. UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25:1605–1612. 47. Guex N, Peitsch MC. 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18: 2714 –2723. 48. Emsley P, Cowtan K. 2004. Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60:2126 –2132. 49. Emsley P, Lohkamp B, Scott WG, Cowtan K. 2010. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66:486 –501. 50. Rossmann MG, Arnold E, Erickson JW, Frankenberger EA, Griffith JP, Hecht Johnson H-JJE, Kamer G, Luo M, Mosser AG, Rueckert RR, Sherry B, Vriend G. 1985. Structure of a human common cold virus and functional relationship to other picornaviruses. Nature 317:145–153. 51. Lin J, Lee LY, Roivainen M, Filman DJ, Hogle JM, Belnap DM. 2012. Structure of the Fab-labeled “breathing” state of native poliovirus. J. Virol. 86:5959 –5962. 52. Lin J, Cheng N, Chow M, Filman DJ, Steven AC, Hogle JM, Belnap DM. 2011. An externalized polypeptide partitions between two distinct sites on genome-released poliovirus particles. J. Virol. 85:9974 –9983. 53. Hewat EA, Neumann E, Blaas D. 2002. The concerted conformational changes during human rhinovirus 2 uncoating. Mol. Cell 10:317–326. 54. Hewat EA, Blaas D. 2004. Cryoelectron microscopy analysis of the structural changes associated with human rhinovirus type 14 uncoating. J. Virol. 78:2935–2942. 55. Curry S, Chow M, Hogle JM. 1996. The poliovirus 135S particle is infectious. J. Virol. 70:7125–7131.
Journal of Virology
Downloaded from http://jvi.asm.org/ on June 12, 2014 by guest
26. Hadfield AT, Lee WM, Zhao R, Oliveira MA, Minor I, Rueckert RR, Rossmann MG. 1997. The refined structure of human rhinovirus 16 at 2.15 A resolution: implications for the viral life cycle. Structure. 5:427– 441. 27. Rossmann MG, Bella J, Kolatkar PR, He Y, Wimmer E, Kuhn RJ, Baker TS. 2000. Cell recognition and entry by rhino- and enteroviruses. Virology 269:239 –247. 28. Rueckert RR. 1996. Picornaviridae: the viruses and their replication, p 609 – 654. In Fields BN, Knipe DM, Howley PM (ed), Fields virology (3rd ed). Lippincott-Raven, Philadelphia, PA. 29. Belnap DM, Filman DJ, Trus BL, Cheng N, Booy FP, Conway JF, Curry S, Hiremath CN, Tsang SK, Steven AC, Hogle JM. 2000. Molecular tectonic model of virus structural transitions: the putative cell entry states of poliovirus. J. Virol. 74:1342–1354. 30. Bubeck D, Filman DJ, Cheng N, Steven AC, Hogle JM, Belnap DM. 2005. The structure of the poliovirus 135S cell entry intermediate at 10angstrom resolution reveals the location of an externalized polypeptide that binds to membranes. J. Virol. 79:7745–7755. 31. Levy HC, Bostina M, Filman DJ, Hogle JM. 2010. Catching a virus in the act of RNA release: a novel poliovirus uncoating intermediate characterized by cryo-electron microscopy. J. Virol. 84:4426 – 4441. 32. Bostina M, Levy H, Filman DJ, Hogle JM. 2011. Poliovirus RNA is released from the capsid near a twofold symmetry axis. J. Virol. 85:776 – 783. 33. Garriga D, Pickl-Herk A, Luque D, Wruss J, Caston JR, Blaas D, Verdaguer N. 2012. Insights into minor group rhinovirus uncoating: the X-ray structure of the HRV2 empty capsid. PLoS Pathog. 8:e1002473. doi: 10.1371/journal.ppat.1002473. 34. Wang X, Peng W, Ren J, Hu Z, Xu J, Lou Z, Li X, Yin W, Shen X, Porta C, Walter TS, Evans G, Axford D, Owen R, Rowlands DJ, Wang J, Stuart DI, Fry EE, Rao Z. 2012. A sensor-adaptor mechanism for enterovirus uncoating from structures of EV71. Nat. Struct. Mol. Biol. 19:424 – 429. 35. Bubeck D, Filman DJ, Hogle JM. 2005. Cryo-electron microscopy reconstruction of a poliovirus-receptor-membrane complex. Nat. Struct. Mol. Biol. 12:615– 618. 36. Bostina M, Bubeck D, Schwartz C, Nicastro D, Filman DJ, Hogle JM. 2007. Single particle cryoelectron tomography characterization of the structure and structural variability of poliovirus-receptor-membrane complex at 30 A resolution. J. Struct. Biol. 160(2):200 –210. 37. Yeates TO, Jacobson DH, Martin A, Wychowski C, Girard M, Filman DJ, Hogle JM. 1991. Three-dimensional structure of a mouse-adapted type 2/type 1 poliovirus chimera. EMBO J. 10:2331–2341. 38. White J, Helenius A. 1980. pH-dependent fusion between the Semliki forest virus membrane and liposomes. Proc. Natl. Acad. Sci. U. S. A. 77: 3273–3277.
